Optimal Pairs Trading Strategies with Flexible Position Sizing and Stop-Loss Boundaries
Abstract:
Our study aims to investigate the profitability of an optimized mean reversion trading strategy in the context of the US equity market. In contrast to conventional pair trading strategies, a thorough method that combines time series, stochastic control techniques, and cointegration is utilized to build the best static pair trading portfolio for the Ornstein-Uhlenbeck process, with parameters that are precisely calculated. Initial assessment in- volves a comprehensive evaluation of multiple metrics to ensure the selected pairs exhibit pre-trade mean reversion characteristics. Subsequently, Ornstein Uhlen beck process parameters are finely tuned to address the varying degrees of stationarity observed in different spread scenarios. Dynamic contrarian trading signals are then derived from model parameters, with thresholds and in-sample period lengths optimized through it erative testing. Analysis of historical data pertaining to five pairs demonstrates that the proposed pair trading strategy outperforms traditional cointegrated pairs, yielding higher returns both within and beyond sample periods, with an average excess annualized return exceeding 8%. Notably, the strategy’s adaptability is high lighted by the dynamic adjustment of model parameters and trading strategies over time, including position sizing, directional bias, and stop- loss thresholds, thereby enhancing robustness and adaptability. Further more, validation of the model’s ability to swiftly adjust portfolios in response to high-risk events validates its effective- ness in mitigating risks while maximizing returns.
